intelligence @ biology, algorithms, AI

interesting research, and it get quite strange, unexpected

apparently "intelligence" is not just "emerging phenomenon", it is a property of all things, 
including such abstract things like computer algorithms, and AI LLM models

#486 – Michael Levin: Hidden Reality of Alien Intelligence & Biological Life | Lex Fridman Podcast

Michael Levin is a biologist at Tufts University working on novel ways to understand and control complex pattern formation in biological systems.

Transcript for Michael Levin: Hidden Reality of Alien Intelligence & Biological Life | Lex Fridman Podcast #486 - Lex Fridman

Here is AI (Gemini 3 Pro) summary of transcript:

In this conversation, Michael Levin argues that intelligence exists on a continuum defined by a "cognitive light cone," requiring a new framework to recognize and communicate with unconventional minds in biological and artificial systems.

He details his research on Xenobots and Anthrobots, demonstrating how bioelectric networks act as a reprogrammable software layer that guides cells to form novel organisms and heal tissue without genetic modification.

Levin explores the "Platonic space" hypothesis, suggesting that physical brains and even simple algorithms function as "thin client" interfaces tapping into pre-existing, universal patterns of intelligence.

The discussion concludes with the implications of these theories for advancing regenerative medicine, understanding the intrinsic motivations of AI, and potentially reversing biological aging.

Michael Levin: Hidden Reality of Alien Intelligence & Biological Life | Lex Fridman Podcast #486 - YouTube

• 2:03:48 – Unexpected intelligence of sorting algorithms

1. How Sorting Algorithms Deal with Unreliable Hardware "Intelligently"

Michael Levin describes an experiment involving standard sorting algorithms (like Bubble Sort) to demonstrate that even simple code can exhibit unexpected "competencies" or goal-directed behavior when faced with barriers.

• The Experiment: Levin’s team took a standard sorting algorithm designed to order a list of numbers. They introduced "unreliable hardware" by arbitrarily "breaking" one of the numbers so that it refused to move when the algorithm tried to swap it.

• The Result: Crucially, they did not change the code to handle this error. They simply ran the standard algorithm on this "broken" hardware.

• The "Intelligent" Behavior: Instead of failing completely or halting, the algorithm managed to sort the rest of the list by moving the other numbers around the immovable block.

• Interpretation (Delayed Gratification): Levin argues this mimics delayed gratification. In a standard sort, every move usually increases "sortedness." However, when the algorithm hits the immovable number, it has to temporarily decrease the overall sortedness (making the list more disordered) to maneuver other numbers around the blockage. The system "accepts" a temporary setback to achieve the final goal.

• Implication: Levin claims this shows that goal-directedness (agency) isn't necessarily something magical added by complex brains. Even simple algorithms have a "competency" to traverse a problem space and navigate around barriers—capabilities that were not explicitly programmed but exist in the "space between" the code’s instructions.

2. Intelligence as an "Interface" vs. Emerging Phenomenon

Levin explicitly challenges the standard materialist view that intelligence and consciousness are solely "emergent" properties generated by the brain's complexity.

• The "Thin Client" Hypothesis: Levin proposes that the brain (and other biological or computational systems) acts as a "thin client" or interface. In computing, a "thin client" (like a web browser) doesn't do the heavy processing itself; it connects to a massive server that handles the work.

• Platonic Space of Minds: He suggests there is a "Platonic space" of mathematical and cognitive patterns. Just as mathematical truths (like the distribution of prime numbers) exist independently of us discovering them, Levin believes behavioral patterns and rudimentary forms of mind also pre-exist in this abstract space.

• Rejection of "Emergence": He criticizes the term "emergence" as often being a label for "we were surprised this happened." He argues it is scientifically more productive to view these patterns as pre-existing potentials that physical systems "pull down" or "ingress."

• The Radio Analogy: He uses the analogy of a radio or TV. If you damage a TV, the picture changes, but that doesn't mean the TV creates the movie. Similarly, damaging the brain affects consciousness, but that doesn't prove the brain generates it—it only proves the brain is the necessary receiver/interface for it.

• Universal Access: This implies that intelligence is not unique to biology. If you build the right "interface" (whether it's a xenobot, a biological brain, or an AI), you can tap into these universal cognitive patterns. The "software" of intelligence exists effectively everywhere, waiting for the right hardware to run it.